scrapegraphai/scrapegraph-100k-finetuning

该数据集是[ScrapeGraphAI-100k](https://huggingface.co/datasets/scrapegraphai/scrapegraphai-100k)数据集的微调就绪版本，包含25,244个训练样本和2,808个测试样本。这些样本是经过筛选的结构化提取示例，其中大型语言模型(LLM)根据用户定义的JSON模式从网页内容中提取数据。数据集经过预处理，包括过滤超出字符限制的示例、将长内容分块以及按90/10的比例分割训练和测试集。

This dataset is a finetuning-ready version of the [ScrapeGraphAI-100k](https://huggingface.co/datasets/scrapegraphai/scrapegraphai-100k) dataset. It contains 25,244 training and 2,808 test curated structured extraction examples where an LLM extracts JSON data from web content according to a user-defined schema. The dataset was preprocessed from the raw 100k collection by filtering examples exceeding character limits, chunking long content into ~4096-token windows with 128-token overlap, and splitting into train/test (90/10) with a fixed seed for reproducibility.